Download Bohr`s Complementarity and Kant`s Epistemology

Bohr, 1913-2013, Séminaire Poincaré XVII (2013) 145 – 166
Séminaire Poincaré
Bohr’s Complementarity and Kant’s Epistemology
Michel Bitbol
Archives Husserl
ENS - CNRS
45, rue d’Ulm
75005 Paris, France
Stefano Osnaghi
ICI - Berlin
Christinenstraße 18-19
10119, Berlin, Germany
Abstract. We point out and analyze some striking analogies between Kant’s
transcendental method in philosophy and Bohr’s approach of the fundamental
issues raised by quantum mechanics. We argue in particular that some of the
most controversial aspects of Bohr’s views, as well as the philosophical concerns
that led him to endorse such views, can naturally be understood along the lines
of Kant’s celebrated ‘Copernican’ revolution in epistemology.
1
Introduction
Contrary to received wisdom, Bohr’s views on quantum mechanics did not gain universal acceptance among physicists, even during the heyday of the so-called ‘Copenhagen interpretation’ (spanning approximately between 1927 and 1952). The ‘orthodox’ approach, generally referred to as ‘the Copenhagen interpretation’, was in fact
a mixture of elements borrowed from Heisenberg, Dirac, and von Neumann, with a
few words quoted from Bohr and due reverence for his pioneering work, but with no
unconditional allegiance to his ideas [Howard2004][Camilleri2009]. Bohr’s physical
insight was, of course, never overtly put into question. Yet many of his colleagues
found his reflections about the epistemological status of theoretical schemes, as well
as his considerations on the limits of the representations employed by science, obscure and of little practical moment – in a word: too philosophical.1 In addition, it
proved somehow uneasy to reach definite conclusions as to the true nature of this
philosophy.
Although Bohr’s views have occasionally been criticized for their alleged positivistic leaning,2 a number of commentators have more aptly pointed out their
pragmatist aspects, especially in view of Bohr’s apparently operationalist definition of microphysical phenomena, and of his documented interest for William James
[Stapp1972][Murdoch1987]. We know that Bohr had a lifelong friendship with the
Danish philosopher Harald Høffding (who was a close friend of his father, and
taught a course at the university of Copenhagen, which Bohr attended as a student)
1 See,
e.g., Holton [1978, p.162], Osnaghi [2009, pp.101-2].
See also the discussion in Howard [2004] and Faye [2009].
2 [Bunge1955].
146
M. Bitbol and S. Osnaghi
Séminaire Poincaré
[Faye1991]. Through Høffding, Bohr was exposed to Kantian influences (partly reshaped by Høffding’s own rather pragmatist views). Even though Bohr himself did
not bother to pinpoint his debt towards Kant, such influences are arguably responsible for the Kantian pattern that can be discerned amidst Bohr’s characteristic
mixture of philosophical styles, and more specifically in his epistemological reflection on quantum mechanics.
Unsurprisingly, therefore, early neo-Kantian analyses of quantum mechanics,
such as Carl-Friedrich von Weizsäcker’s [Heisenberg1971, Ch. X], Grete Hermann’s
[1996], and Ernst Cassirer’s [1956], took advantage of Bohr’s position, instead of
distantiating themselves from it. The Kantian element in Bohr’s thought was subsequently recognized by a number of historians and philosophers of physics,3 while
being minimized or denied by others.4 The reasons for such disagreement are easy
to understand: on the one hand, the Kantian component of Bohr’s epistemology
is buried in his idiosyncratic remarks about the role of measuring devices and the
boundaries of theoretical domains; on the other hand, after more than three centuries of erudite commentaries, the gist of Kant’s philosophy remains a matter of
debate. Any attempt to bring out the affinities and differences between Bohr and
Kant can, therefore, hardly avoid engaging in a twofold process of clarification.5
In the present paper, we will in particular be concerned with showing that some
of Bohr’s crucial moves, while seemingly conflicting with the letter of Kant’s work,
are in fact very much in tune with its spirit. The first section sorts out the various
dimensions of Kant’s theory of knowledge, and identifies what was retained and what
was left out by Bohr. We insist in particular on the striking analogy between, on
the one hand, Kant’s strategy to ‘disentangle’ the object and the subject from the
cognitive relation which lies at the heart of our experience, and, on the other hand,
Bohr’s endeavor to come to grips with the ‘indivisibility’ of quantum phenomena. In
the subsequent sections we analyze in turn Bohr’s ‘principal distinction’ between the
measuring instrument and the measured object (which we relate to Kant’s general
reflection upon the subjective conditions of possibility of knowledge), Bohr’s concept
of complementarity (which we relate to Kant’s theory of the object), and the peculiar
role of classical concepts within Bohr’s approach (which we relate to the a priori
forms that, according to Kant, enable the subject to constitute the objectivity of
knowledge).
2
Kant’s and Bohr’s ‘Copernican’ turns
As a way to characterize what he regarded as a radical rupture with previous philosophical traditions, Kant described his epistemology as a ‘Copernican revolution’.
By this term, Kant meant to emphasize the analogy between his critical approach
and Copernicus’ decision to base the explanation of the apparent motion of the planets on the relations between the motion of the Earth and the orbit of the planets,
instead of sticking to a theory of their intrinsic kinematics. Rather than remaining
exclusively fascinated by his object of study (the planets), Copernicus focused on
the situation of the astronomer on planet Earth, and addressed the problem of how
this particular situation contributes to shaping cosmological knowledge. The gener3 [Hooker1972],
[Honner1982], [Murdoch1987], [Chevalley1991], [Kaiser1992], [Brock2003].
[Pais1991].
5 [Held1995], [Pringe2007], [Cuffaro2010], [Kauark-Leite2012].
4 [Folse1978,1985],
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
147
alization of this reflective stance to scientific knowledge in general is described by
Kant in a famous passage:
Thus far it has been assumed that all our cognition must conform to
objects. Let us try to find out by experiment whether we shall not make
better progress, if we assume that objects must conform to our cognition.
[Kant1996, BXVI, p.21]
This does not mean that objects are so to speak created by our cognition, but
(i) that one cannot dispense with a proper analysis of our own faculties of knowing, if knowledge is to be understood at all, and (ii) that the form of objects is
predetermined by a set of cognitive conditions enabling us to overcome the variegation of fleeting subjective appearances, and to circumscribe some invariant phenomena which can be intersubjectively recognized and designated. The term ‘object’
is accordingly understood as referring to such experiential invariants, rather than
to something beyond experience. Kant then undertook to spell out what, in the
structure of our cognition, makes the identification of unified invariant phenomenal patterns possible. He found two classes of such structures. The first one is the
continuum of our sensory givenness, namely space and time. The second one is the
table of the general concepts of our understanding (or categories), which are used
to bring the manifold of sensory appearances under a common organization. Among
the latter, we find: (i) the category of substance, which permanently unifies a set of
attributes, and (ii) the category of causality, which enables us to differentiate between unruly subjective successions and law-like sequences of phenomena that any
subject can identify.
Like Kant, Bohr was thoroughly concerned with the inherent structures of our
cognitive apparatus, and he thought that we cannot dispense with studying them,
if we want to make sense of the objectivity of scientific knowledge. Bohr’s notion
of experience is definitely Kantian in that he takes ‘the boundary of our concepts’
to be ‘exactly congruent with the boundary to our possibilities of observation.’6
According to Bohr, ‘all knowledge presents itself within a conceptual framework ’,
where, by ‘a conceptual framework’, Bohr means ‘an unambiguous logical representation of relations between experiences.’ [Bohr1934, pp.67-8, our emphasis] In
his reflective analysis of the structure of our capability to know, however, Bohr did
not address, as Kant did, mental faculties such as sensibility and understanding.
Instead, he focused on a technological counterpart of sensibility, namely the measuring apparatus, and on an intersubjective counterpart of understanding, that is,
language. On the one hand, he foregrounded the link between the phenomena and
the experimental context in which they occur. On the other hand, partly following
(and even anticipating) the linguistic turn of the philosophy of his time, he saw
in the conditions for unambiguous communication an essential feature of scientific
objectivity [Bitbol1996b, pp.263-9].
Bohr’s ‘Copernican turn’ can naturally be understood in the context of the
quantum revolution that took place in the first decades of the twentieth century. As
long as a scientific paradigm [Kuhn1962] is generally accepted, science can be taken
to pursue an increasingly precise characterization of its purported objects. However,
when the paradigm is crumbling, the ontological status of these very objects becomes
suspicious, and the standard procedures for extracting invariant phenomena are also
6 Niels
Bohr, letter to Albert Einstein, 13 April 1927, quoted in Honner [1982, p.7].
148
M. Bitbol and S. Osnaghi
Séminaire Poincaré
put into question. One then falls down onto the only firm ground left, which is
‘ordinary’ experience along with the results of experimental inquiry, in so far as
the latter have recognizable implications within the former. Bohr’s view that ‘the
task of science is both to extend the range of our experience and to reduce it to
order’ [Bohr1934, p.1] enables him straightforwardly to deal with such a situation.
This instrumentalist attitude (which was all but uncommon among the physicists
involved in the creation of quantum mechanics) is however decisively supplemented,
in Bohr’s case, by a careful analysis of the conditions of possibility of experience.
He is, in other words, concerned with precisely the sort of reflective knowledge
that Kant called transcendental, and regarded as the fundamental subject-matter of
‘post-Copernican’ metaphysics.
Like the more common term ‘transcendent’, the adjective ‘transcendental’ applies to something that ‘exceeds experience’. This ‘excess’, however, can be realized
in two antithetical ways. A transcendent object exceeds experience in so far as it
is said to exist beyond experience, as a remote (and intellectually reconstructed)
external cause of experienced phenomena. Conversely, a transcendental structure
exceeds experience because it is a precondition of experience: it shapes experience
without being part of experience. Moreover, as long as the act of knowing develops,
such a transcendental structure is bound to remain in the silent background of this
act. We bump here into the extraterritorial status of the precondition of knowledge
- a status that Kant, who created the very concept of a transcendental epistemology,
regarded as a strong logical requirement. In his own terms, the question of how the
condition of possibility of knowledge ‘. . . is (itself) possible, will not admit of any
further solution or answer, because we invariably require [such a condition] for all
answers and for all thought of objects.’ [Kant1994, §36] In other words, what plays
the role of the knower cannot be known in the very process of knowing.
As we will see in Section 3, Bohr held very similar views with regard to measurement. In particular, he took the attempts to include the act of observation of a
quantum phenomenon in the description of the phenomenon itself, to be fundamentally misguided. Echoing Kant, we could say that what preconditions the possibility
of a quantum description cannot be described quantum-mechanically in the very
process of describing. This is not to deny that quantum mechanics, as one of the
most accomplished realizations of the ideal of universal description pursued by the
natural sciences, could indeed describe any phenomena. Yet, in doing so, it could not
avoid leaving the preconditions for description outside its scope. As a well-known
article about the measurement problem of quantum mechanics puts it: the quantum theory can describe anything, but not everything [Peres1982][Fuchs2000]. Bohr’s
argument for such a conclusion is that the ‘indivisibility of the quantum of action’
precludes the description of the measurement interaction [Murdoch1987, Ch. 5]. This
is so because, if the measuring instrument is to serve as a measuring instrument, at
least part of it cannot obey the uncertainty relations, and this precludes its representation qua measuring instrument within the quantum-mechanical account of the
phenomenon under study. Instead, the instrument is to be identified with the conditions that make it possible to observe and to describe the phenomenon in the first
place. As such, it is placed ‘in the background’, and referred to using ‘common language supplemented with the terminology of classical physics.’ [Bohr1998, pp.142-3]
This implies that a ‘cut’ must be introduced somewhere in our mental picture of
the measuring chain, in order to separate the object-system to be described by the
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
149
quantum symbolism from the measuring instrument, which is described in classical
terms. Characteristically, in Bohr’s writings, this issue is often related to what he
calls the ‘subject-object separation’ [Bohr1963, p.12].
Our analysis of the term ‘transcendental’ has thus far focused on the subjective
term of the cognitive relation. Let us now turn to the other, objective term of
the relation. Beyond the horizon of experience, in the region denoted by the term
‘transcendent’, lies, according to Kant, the ideally conceived ‘thing in itself’, that
is, the thing as it is independently of any relation with our faculties of knowledge.
Things in themselves are to be contrasted with the objects of experience, which,
according to Kant, are intrinsically relational:
Things . . . are given in intuition with determinations that express mere
relations without being based on anything intrinsic; for such things are not
things in themselves, but are merely appearances. Whatever [characteristics] we are acquainted with in matter are nothing but relations (what we
call its intrinsic determinations is intrinsic only comparatively); but among
these relations there are independent and permanent ones, through which
a determinate object is given to us. [Kant1996, B341, p.340]
What we call the ‘properties of material objects’ are only the expression of the
cognitive relations that we establish with our environment; they are not ‘proper’ to
some object, but rather arise as an unanalyzable byproduct of our interaction with
‘it’ (the quotation marks surrounding the word ‘it’ are justified by the fact that, in
Kant’s approach, there are no such things as preconstituted objects placed before
a passive sensorial or experimental apparatus). What exists beyond these cognitive
relations and independently of them is in principle unreachable, since reaching it
would precisely mean establishing a cognitive relation with it. So much so that
some commentators concluded that, by the term ‘thing in itself’, Kant merely refers
to the impossibility of disentangling ourselves completely from the content of our
knowledge, thereby drawing a sharp distinction between the bulk of what we know
and our own contribution qua knowers.7
Here too a close analogy with Bohr can be drawn. Starting from the observation
that ‘the properties of atoms are always obtained by observing their reactions under
collisions or under the influence of radiation’, Bohr showed that the quantum of
action compels us to acknowledge the existence of a fundamental ‘limitation on the
possibilities of measurement’ [Bohr1934, p.95]. Unlike Heisenberg and others, however, Bohr did not endorse the interpretation according to which the ‘disturbance’
introduced by the measuring agent prevents us from having complete knowledge
of the properties of the object under study. Nor did he suggest that we should regard those properties as intrinsically fuzzy or unsharp. Rather, he pointed out that,
given the ‘impossibility of a strict separation of phenomena and means of observation’ (Ibid., p.96), the very notion of attributes that would be proper to the atomic
object becomes unworkable. Since ‘. . . interaction forms an inseparable part of the
phenomena’ [Bohr1963, p.4], any discourse about phenomena going on in nature
independently of any measuring interaction appears to be meaningless. In his later
writings, Bohr took a further step: he advocated the ‘interactionality conception of
microphysical attributes’, according to which properties can only be meaningfully
7 [Hintikka1991].
See also the discussion in Allison [2004, Ch.3].
150
M. Bitbol and S. Osnaghi
Séminaire Poincaré
defined in certain experimental contexts, and not in others [Jammer1974, p.160].8
The consequences of Bohr’s inseparability thesis are exactly the same as those
of Kant’s thorough relationism. Like the latter, the former leaves us two options:
either (i) accepting that there is something like a ‘micro-object in itself’ that we can
know only obliquely, by means of successive interactive approaches; or (ii) declaring that any term referring to some such obliquely knowable ‘micro-object in itself’
is a fake name for the impossibility of breaking up the wholeness of the phenomena. Arguably, Bohr’s concept of complementarity (which we discuss in Section 4)
was, at least to some extent, intended to make option (i) viable. Like the ‘cut’ that
keeps the measuring instrument separated from the phenomenon under study, also
complementarity is meant as a conceptual tool for extracting objective results from
the quantum symbolism. Bohr takes it to be essential for that purpose that we can
refer the measurement results to the properties or the behaviour of some object.
And since it is only within a specific set of experimental situations (in which compatible observables are measured) that certain classes of predicative sentences can
be employed without inconsistency, Bohr restricts his definition of ‘phenomenon’ to
‘observations obtained under specified circumstances, including an account of the
whole experimental arrangement’ [Bohr1958, p. 64]. The task of complementarity is
then to bring phenomena occurring in incompatible situations together, as if they
referred to the same object.
The last feature of Kant’s epistemology that we want to consider is the organic
articulation of what he regarded as the two sources of knowledge, namely sensibility
and understanding. Since Kant considered that the unity brought by concepts could
only concern the data of sensory intuition, he looked for a proper locus of connection
between the forms of understanding and the contents of intuition, which he found
in what he termed the ‘schematism of pure imagination’ [Kant1996, B177, p.210].
Schematism can be conceived of as the ability to devise pictures that guide our
possible actions in so far as such actions purport to anticipate certain contents of
sensory intuition in a systematic way. An example of such a connection is causality.
As a concept of pure understanding, causality applies, according to Kant, to sensory
contents pre-ordered by a spatiotemporal structure. And its application is mediated
by the scheme of succession according to a rule (a generic term for the trajectories of
a dynamics), which allows one to anticipate later phenomena based on the knowledge
of an appropriate set of earlier ones.
It is precisely at this point that Bohr parts company with Kant. Being faced
with the puzzling phenomena of atomic physics, Bohr became increasingly diffident
of some components of Kant’s account of the constitution of experience. At an early
stage of the development of quantum mechanics, he questioned the universal applicability of the category of causality, deeming that it might preclude the possibility of
providing a ‘pictorial’ spatiotemporal description of the trajectories (see Section 5).
Subsequently, he endorsed the view that no single ‘picture’ could be meaningfully
used to represent atomic processes: pictures were to be viewed as purely ‘symbolic’,
not to say ‘poetical’, devices for accounting of the atomic processes [Heisenberg1971,
Ch. 3]. Eventually, Bohr came to regard the category of causality and the spatiotemporal coordination of phenomena as mutually exclusive (see Section 4). It was, at this
point, no longer possible to understand the anticipation of phenomena on the basis
8 See Murdoch [1987, Ch. 7]. For a discussion of the analogy between Bohr’s and Kant’s respective notions of
phenomenon, see Kaiser [1991].
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
151
of a continuous spatiotemporal representation. Anticipation was rather achieved by
means of ‘. . . a purely symbolic scheme permitting only predictions . . . as to results
obtainable under conditions specified by means of classical concepts.’ [Bohr1963,
p.40] Thus, a major keystone of Kant’s theory of knowledge had been removed, and,
to many physicists of the time, the entire building looked like it was doomed to
crumbling.
The point of disagreement between Bohr and Kant can be characterized in a
few words. Kant claimed that the forms of sensibility and understanding he had
identified were a priori conditions of possibility of objective knowledge in general.
This suggested that these forms, as well as their articulation, were fixed for ever by
way of necessity, and that no scientific knowledge might be conceived which would
not fit within such a scheme. As for Bohr, he accepted the typically Kantian idea of
‘subjective forms’ that constitute experience. He stressed that ‘. . . in spite of their
limitations, we can by no means dispense with those forms of perception which
colour our whole language and in terms of which all experience must ultimately be
expressed.’ [Bohr1934, p.5] But Bohr also considered that, in view of these limitations, ‘. . . we must always be prepared to expect alterations in the points of views
best suited for the ordering of our experience.’ (Ibid., p.1) The idea of modifying the
so-called ‘a priori ’ forms of human knowledge according to the advances of scientific
research was so averse to Kant, that this alone sufficed for some philosophers of
science to rule out any deep similarity between Bohr’s and Kant’s respective theories of knowledge [Folse1978]. However, even in this particular respect, the relation
between the two theories is more complex than it at first sight appears.
To begin with, Kant’s epistemology need not be viewed as the static articulation
of the ideas displayed in Kant’s own system. One can also regard it as a research
program. Such a program was developed by an entire school of neo-Kantian philosophers, whose major move was to historicize and relativize the a priori forms of
knowledge,9 by suggesting that their function (namely, to unify, and extract invariants from, the manifold of appearances) could be ascribed, for instance, to plastic
‘symbolic forms’ [Cassirer1965] or to historically relative ‘principles of coordination’
[Reichenbach1965]. Philosophical pragmatism itself can, to some extent, be traced
to a Kantian framework, provided that the latter’s emphasis on the absolute a priori
is attenuated [Putnam1995]. These examples show that Kant’s epistemology, when
conceived as a research program, might indeed prove flexible enough to account for
quantum knowledge.
Conversely, one might argue that Bohr’s epistemology is itself characterized by
a sort of a priori component, namely ‘classical concepts’. As we have seen, Bohr’s
account of quantum phenomena relies on classical concepts in two different contexts:
the description of the measuring apparatus and that of the complementary features
of a given atomic ‘object’. This is no accident. Bohr was always sceptical towards
the idea that atomic phenomena demanded the dismissal of the classical conceptual
framework. In his reply to a letter in which Schrödinger urged ‘the introduction of
new concepts’ (though, as he said, this would entail a ‘re-organisation involv[ing] the
most profound levels of our knowledge, space, time and causality’), Bohr remarked:
‘I am scarcely in complete agreement with your stress on the necessity of developing
‘new’ concepts. . . . The ‘old’ experimental concepts seem to me to be inseparably
9 [Friedman1992],
[Bitbol1998], [Pradelle2013].
152
M. Bitbol and S. Osnaghi
Séminaire Poincaré
connected with the foundation of man’s powers of visualizing.’ 10 He reiterated the
same point in his writings, arguing that it was ‘. . . [un]likely that the fundamental
concepts of the classical theories will ever become superfluous for the description of
physical experience.’ [Bohr1934, p. 16, our emphasis]
The need to retain classical concepts can be generically related, as Bohr repeatedly did, to the conditions of possibility of unambiguous communication. As we will
see in Section 5, however, Bohr’s lack of enthusiasm for any project involving the
replacement of classical concepts stems less from his attachment to those particular
‘forms of perception’ than from the conviction that what quantum mechanics demands is not so much a new set of concepts as a deeper understanding of the way
concepts in general enter the construction of objective knowledge. This, once again,
sounds like a distinctly Kantian concern. Bohr’s suggestion that we should think of
‘the viewpoint of complementary forms’ as ‘a consistent generalization of the ideal
of causality’ [Bohr1958, p.27, our emphasis] can be understood along these lines.
In so far as we confine our analysis to the measuring apparatuses and other
massive objects as considered within ordinary experience, Bohr thinks that we are
allowed to rely on ‘our accustomed forms of perceptions’, in spite of having learned
from quantum mechanics that these are mere ‘idealizations’ (Ibid., p.5). In this sense,
Bohr can be thought to acknowledge the enduring pragmatic value of Kant’s a priori
forms. The empirical domain to which such forms apply is nevertheless severely
hampered [Bitbol2010]. It does not extend to the whole field of microphysics, but
only to ordinary experiences. Kant’s a priori forms are taken as an anthropological
condition of possibility for the technological conditions of possibility of microphysical
research. Therefore, as Heisenberg pointed out, they can be viewed as a second-order
condition of possibility of microphysical knowledge: ‘What Kant had not foreseen
was that these a priori concepts can be the conditions for science and at the same
time have a limited range of applicability.’ [Heisenberg1990, p.78] The forms that
directly precondition our familiar experience and the phenomena of classical physics
have a limited range of applicability; yet they indirectly precondition all empirical
knowledge beyond that range. This may be seen as Bohr’s simultaneous vindication
and confinement of Kant’s a priori.
3
The ‘agency of measurement’ and the cut
By stipulating that we should use a classical mode of description to account for
the measuring instruments, the measurement outcomes, and the experimental procedures, Bohr seems to grant them a sort of ‘extraterritorial status’. It is important
to realize, however, that Bohr’s prescription in no way presupposes or implies an
ontological distinction between macroscopic and microscopic systems. There is nothing in the physical nature of macroscopic objects that distinguishes them from the
microscopic ones, and which rules out the possibility of describing them as quantum systems.11 Bohr’s concern is rather to emphasize the specific function that
the measuring apparatuses accomplish in the system of knowledge: that of ensuring
10 Erwin Schrödinger, letter to Niels Bohr, 5 May 1928, and Bohr’s reply, 23 May 1928, both quoted in Murdoch
[1987, p. 101].
11 This point is often overlooked. Rovelli [1996, p.1671] remarks for example that ‘the disturbing aspect of Bohr’s
view is the inapplicability of quantum theory to macrophysics’, and he understands such a view as implying that
‘the classical world is physically distinct from the microsystems’. See also, e.g., Zurek [2003], Weinberg [2005], as
well as Hugh Everett’s analysis of Bohr’s position in Osnaghi [2009].
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
153
the intersubjective agreement about experimental results and procedures, thereby
fulfilling a condition of possibility of objective experience. The functional status of
the ‘principal distinction’ between measuring instruments and object-systems [Bohr,
1998, p.81] is confirmed by Bohr’s emphasis on the fact that the boundary between
the classical and the quantum domain is by no means fixed, and we are left the
‘free choice’ to locate it at one or the other point of the measuring chain [Bohr1998,
pp.73-82]. By choosing to describe the measuring apparatus in classical terms, Bohr
cuts the measurement problem off at its root [Murdoch1987, p.114]. On the one
hand, in so far as a classic-like actualistic description of the relevant part of the
measuring instruments is taken as the fundamental presupposition of any quantum
account of the phenomena, there is no need to figure out a mechanism for the transition between potentialities and actuality. On the other hand, since the quantum
description is prevented from extending to the totality of the measurement chain,
there is no such thing as a superposition of pointer macrostates.
While dissolving the traditional measurement problem, Bohr’s strategy is nevertheless faced with an issue of consistency. The same system can alternatively be
seen as a measuring instrument (in which case it is described classically) or as an
object-system (in which case the quantum symbolism applies). As Bohr hinted, the
issue can in principle be fixed by thermodynamic considerations, in so far as the
macroscopic number of degrees of freedom of real measuring apparatuses is taken
into account [Daneri1962][Rosenfeld1965]. Although no unitary process can bring
the quantum account exactly to coincide with a classical distribution of definite
pointer positions, one may argue that the two become practically indistinguishable
soon after the measurement has taken place. Despite its practical effectiveness and
arguable consistency, saying that Bohr’s approach is not very popular among physicists would be an understatement.12 For, by preventing in principle the quantum
theory from accounting (at least retrospectively) for the experimental conditions of
its own assessment, it seems to imply that any physical description is fundamentally
incomplete [Omnès1992, pp.340-1][Weinberg2005].
In the late 1950s, the preceding objection was raised and intensely debated
with Bohr and his collaborators by Hugh Everett, then a PhD student at Princeton
under the supervision of John Wheeler [Osnaghi2009]. It is, however, only after
Bohr’s death that the issue started to receive increasing attention. The growing
dissatisfaction with Bohr’s approach culminated in John Bell’s [1990] ‘manifesto’
Against ‘measurement’. In this article, Bell criticized the idea that we should content
ourselves with propositions that are valid only ‘for all practical purposes’, and he
argued that a mature quantum theory should at once get rid of the ‘shifty split’ and
‘refer to the world as a whole’. Elsewhere, Bell [2004] discussed Bohm’s [1952] hidden
variable and Everett’s [1957] ‘relative state’ formulations of quantum mechanics as
two possible ways of ‘completing’ the Copenhagen interpretation along these lines.
More recently, the same program has received a strong impulse by the theoretical and
experimental study of decoherence. As a way to get rid of Bohr’s alleged ‘dualism’,
it has been suggested that it might be sufficient to prove that the classical behavior
of macroscopic systems is predicted by quantum mechanics itself, provided that the
effects of decoherence are taken into account [Zurek2003][Joos2003].
It is interesting to frame the research program just outlined in the more general
philosophical project that aims at the complete ‘naturalization’ of the transcenden12 In
a manuscript of 1955, Everett calls it ‘repugnant’ [Osnaghi2009, p.105].
154
M. Bitbol and S. Osnaghi
Séminaire Poincaré
tal. In so far as the ‘cut’ is supposed to separate the content of knowledge from its
preconditions (which is clearly the function that Bohr attributed to it), any attempt
to generalize the quantum theory for the purpose of doing away with the cut can
be viewed as a step in the direction of providing a fully naturalistic account of the
cognitive process. The guiding idea of such attempts is that, if the measuring apparatus is to be regarded as a natural object (as nothing seems to prevent us from
doing), it should be possible to describe the measurement interaction as a standard
natural process. More generally, it should be possible to consider the whole cognitive operation, including the decision to perform an experiment and the realization
of the occurrence of one particular outcome, as a topic to be addressed by natural
sciences such as neurobiology, biochemistry, optics etc. Wheeler and Everett made
precisely this point during their discussions with the Copenhagen group: ‘Thinking,
experimentation and communication – or psychophysical duplicates thereof – are all
taken by Everett as going on within the model universe.’13
At first sight, naturalization looks like a quite reasonable research program.
Nothing seems to prevent the continuous expansion of the domain of validity of the
sciences of nature towards an increasing disclosure of the cognitive process. Moreover, in so far as one considers classical physics, the indispensability of the pretheoretical level [Mittelstaedt, 1998, pp.8, 104] can go unnoticed, since the pretheoretical
treatment of the measurement process can be made isomorphic to its theoretical description. So much so that isomorphism could in this case be conflated with identity,
which would allows one to understand ‘the measurement process . . . as a special case
of the general laws applying to the entire universe.’ [Bohm1993, p.13] Nothing seems
therefore to hinder the grand project of an entirely naturalized theory of knowledge,
which would render Kant’s intimation of a transcendental epistemology somehow
superfluous.
When it comes to quantum mechanics, however, things are not as simple. Indeed, Bohr’s point can be formulated precisely by saying that the transcendental
approach of knowledge becomes unavoidable when one is concerned with quantum
phenomena. Bohr has two arguments to support this claim. The first, which he
develops at length, especially in his replies to Einstein’s objections, is the abovementioned dynamical argument that in order for measurement to be possible at all
in the quantum domain, we are forced to presuppose that at least part of the instrument does not obey the uncertainty relations. This is a typical Kantian pattern
of reasoning. Bohr assumes that we do have knowledge of atomic phenomena, and
analyzes the conditions that must be satisfied in order for this to be possible. There
is, however, another argument, which recurs in Bohr’s later writings, although he
never takes the trouble of spelling it out in detail. This is the semantic argument according to which the very possibility of communicating the results of an experiment,
as well as the conditions under which these results were obtained, requires that they
be expressed in ordinary language. This may look as a prescription stemming from
a particular theory of meaning, one that privileges ‘observational language’. As we
will see in the last section, however, what really matters for Bohr is that no account
of experience is possible without assuming some conceptual framework (in Bohr’s
sense). Consequently, the attempt to get rid of ‘any presupposition’, by showing that
the assumed conceptual relations emerge ‘spontaneously’ within a suitable account
of experience, can only bring us into a regress. As Bohr’s friend and collaborator
13 John
Wheeler, letter to Alexander Stern, 25 May 1956, quoted in Osnaghi [2009, p. 118].
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
155
Léon Rosenfeld put it in a letter of 1959:
To try (as Everett does) to include the experimental arrangement into
theoretical formalism is perfectly hopeless, since this can only shift, but
never remove, this essential use of unanalyzed concepts which alone makes
the theory intelligible and communicable.14
This conclusion is illustrated by the fact that the shadow of the distinction
between the predictive formalism and the elementary presuppositions needed for
putting it to the test remains visible, and uneliminable, in the naturalizing approaches mentioned above. In the case of Bohm’s theory, the hidden variables whose
distribution is supposed to determine the result of a measurement are made epistemically inaccessible by means of an ad hoc postulate, which might be the shadow
of the background status of the act of measuring. In the case of Everett’s theory,
the pretheoretical level is reflected in the deterministic description of the ‘universal
state vector’. This description is given from the standpoint of an ideal metaobserver,
and is required in order to derive the probabilities recorded in the ‘memories’ of the
naturalized observers who inhabit the various ‘branches’ of the universal state vector. Similar remarks apply to the approaches based on decoherence. Decoherence
can yield quasi-classical probabilistic structures for certain ‘preferred’ observables,
but not a full degree of classicality [Schlosshauer2011]. Moreover, in order to do so,
it must rely on a deliberate selection of the relevant degrees of freedom of the system
within a larger domain of environmental degrees of freedom, which can be seen as
the shadow of the ‘cut’ between the object-system and the experimental context.
4
The ‘measured object’ and complementarity
According to Bohr, a crucial consequence of the fact that the experimental modes of
access to phenomena cannot be separated from the phenomena themselves is complementarity. Let us follow Bohr’s reasoning. In classical physics, where the influence
of the measuring procedure can in principle be substracted from its outcome, the
data obtained by using various instruments ‘. . . supplement each other and can be
combined into a consistent picture of the behaviour of the object under consideration.’ [Bohr1963, p.4] Conversely, in quantum physics, changing the experimental
arrangement is tantamount to changing the holistic phenomenon itself, which therefore turns out to be incompatible with other holistic phenomena. ‘Combination into
a single picture’ of various experimental data then yields contradictions. The contradictions may be overcome by adopting a new, non-pictorial method for articulating
the information derived from various experimental arrangements. Bohr calls this
method ‘complementarity’ because it takes the various pieces of information obtained in mutually exclusive experimental contexts to be jointly indispensable in
order to characterize a given micro-object:
. . . evidence obtained under different experimental conditions cannot
be comprehended within a single picture, but must be regarded as complementary in the sense that only the totality of the phenomena exhausts
the possible information about the objects. [Bohr1958, p.40]
14 Léon
Rosenfeld, letter to Saul Bergmann, 21 December 1959, quoted in Osnaghi [2009, p.117].
156
M. Bitbol and S. Osnaghi
Séminaire Poincaré
Bohr’s complementarity is no simple or unambiguous concept. One may list
three (disputable) applications of the concept of complementarity in quantum mechanics:15
C1 – The complementarity between incompatible variables. ‘In quantum physics
. . . evidence about atomic objects obtained by different experimental arrangements
exhibits a novel kind of complementary relationship.’ [Bohr1963, p.4] A standard
example is the complementarity of the archetypal couple of conjugate variables,
namely position and momentum.
C2 – The complementarity between causation and spatiotemporal location of
phenomena. This was the first explicit formulation of complementarity, which Bohr
stated in his 1927 Como lecture: ‘The very nature of quantum theory. . . forces us to
regard the space-time coordination and the claim of causality, the union of which
characterizes the classical theories, as complementary but exclusive features of the
description, symbolizing the idealization of observation and definition respectively.’
[Bohr1934, pp.54-5]
C3 – The complementarity between the continuous and discontinuous pictures
of atomic phenomena, i.e. between the wave model and the particle model :
The individuality of the elementary electrical corpuscles is forced upon
us by general evidence. Nevertheless, recent experience, above all the discovery of the selective reflection of electrons from metal crystals, requires
the use of the wave theory superposition principles in accordance with the
original ideas of L. de Broglie. . . . In fact, here again we are not dealing
with contradictory but with complementary pictures of the phenomena
which only together offer a natural generalization of the classical mode of
description. (Ibid., p.56.)
C3 is at the same time the most popular and the most controversial version of
complementarity, since it involves remnants of classical representations (as opposed
to classical variables, or classical terminology for the description of measuring apparatuses) [Murdoch1987, p.59]. As Held [1994] points out, after 1935 Bohr ‘. . . tacitly
abandons the idea of wave-particle complementarity’, and starts to regard the concept of complementarity as providing only an indirect ‘clarification’ of the dilemma
of wave-particle dualism [Bohr1963, p.25]. This suggests that the two other formulations of complementarity might in some sense be more fundamental. In the rest of
this section, we will examine each of them in turn.
The C1 version of complementarity, which is likely to be close to the roots of
the concept, is not without difficulties. The most benign difficulty is that the incompatibility of a pair of conjugate variables implied by C1 does not follow from
that of the corresponding experimental arrangements. Indeed, from a narrowly epistemic interpretation of the uncertainty relations, one can derive that our experimental knowledge of the position is incompatible with our experimental knowledge
of the momentum. Yet, this does not necessarily prevent the object from intrinsically ‘possessing’ both properties. As we have seen, however, Bohr’s argument for
the incompatibility of conjugate variables did not rest on a doubtful extrapolation
from epistemic limitations to ontology. His point was rather that the experimental
arrangement is not to be understood as the instrument for revealing the putative
15 [Faye1991],
[Held1994], [Murdoch1987], [Bitbol1996a].
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
157
intrinsic value of a given variable, but rather as an essential part of the definition of
that very variable.
A more serious difficulty bears on the claim that the values of incompatible
variables are jointly indispensable to exhaust knowledge about the object. This
claim seems to conflict with Bohr’s tenet that incompatible variables cannot simultaneously have definite values. As a way to bypass simultaneous possession, one
might argue that joint indispensability refers to the values found when successive
measurements are actually performed. But this solution is not satisfactory either,
because the values can vary according to the order of the measurements. A more
satisfactory alternative is to focus on the observations that are possible when only
the experimental preparation, but not the measurement to be performed, has been
fixed [Held1994]. Along this line of thought, one may reconcile mutual exclusion and
completion of conjugate variables without logical inconsistency: mutual exclusion
pertains to actual experimental arrangements, whereas completion (or exhaustiveness) refers to possible measurements.
An important question arises at this point. Given that complementary variables require incompatible experimental arrangements in order to be measured, are
we compelled to think of them as attributes of one and the same thing? There
is little doubt that, for Bohr, a complete list of conjugate variables provides exhaustive information about an object. This aboutness of experimental information is
stressed again and again: ‘. . . evidence obtained under different experimental conditions . . . must be regarded as complementary in the sense that only the totality
of the phenomena exhausts the possible information about the objects’ [Bohr1958,
p.40]; ‘. . . together (these phenomena) exhaust all definable knowledge about the
objects concerned’ (Ibid., p.90); ‘. . . such phenomena together exhaust all definable
information about the atomic objects’ (Ibid., p.99). The emphasis on aboutness suggests indeed that Bohr conceived of complementarity as a means to provide us with
the possibility of referring to micro-objects as if they were somehow independent
of any experimental procedure – a possibility that would otherwise be precluded
by the unanalyzable (or interactional) nature of quantum phenomena. The atomic
objects Bohr seems to have in mind have the same general predicative structure as
the objects of classical mechanics, although they are not ascribed predicates as such.
Just as classical particles, ‘atomic objects’ are indeed construed by Bohr as points of
convergence of two families of conjugate features such as position and momentum.
Bohr’s account is not without alternatives, however. One might in particular dismiss the presupposition that the experimental outcomes are about objects endowed
with the same predicative structure as the moving bodies of classical mechanics. That
is, one could accept that position and momentum are mutually exclusive, while at
the same time rejecting the idea that they are jointly indispensable for describing
some thing. The interpretation of C1, therefore, places us before a dilemma similar
to that of the ‘object in itself’: should we refer to a ‘micro-object in itself’, characterized by successive, but mutually incompatible, interactive probings; or should
we rather look for a new mode of objectification that retains nothing (not even
the adumbration of a predicative structure) of the classical corpuscularian concept?
If we follow Bohr in adopting the first (conservative) strategy, we must ascribe a
highly non-conventional status to micro-objects. Expressing this status in a Kantian
idiom, we might say that Bohr’s cloudy ‘micro-object’ is a unifying symbol used as
158
M. Bitbol and S. Osnaghi
Séminaire Poincaré
a regulative-heuristic device; it is not a tangible something.16 However, if we adopt
the ‘revolutionary’ strategy of looking for novel forms of objectification, we become
free to construe certain elements of the quantum symbolism as denoting previously
unconceivable objects of knowledge. This is precisely how Schrödinger proposed that
we should interpret the wave function [Bitbol1996a].
Let us finally turn to C2, namely the complementary relation between causality
and the use of space-time concepts. According to Bohr, ‘. . . the use of any arrangement suited to study momentum and energy balance - decisive for the account of
essential properties of atomic systems - implies a renunciation of detailed space-time
coordination of their constituent particles.’ [Bohr1963, p.11] This suggests that C2,
like C1, derive from combining the incompatibility of experimental arrangements
with the interactionality thesis. However, other texts of the earlier period,17 as well
as Heisenberg’s lucid interpretation of Bohr’s position, favour another conception
of the complementarity between causality and space-time coordination, which does
not reduce it to the complementarity of pairs of conjugate variables. In his Physical
principles of the quantum theory, Heisenberg argues that the measurement of any
variable whatsoever involves spatiotemporal aspects. Thus, it is not only space-time
coordination, as any description of phenomena in space-time, which is incompatible with causality. Indeed, one can observe no spatiotemporally circumscribed phenomenon without influencing it in a way that prevents the application of causal laws
[Heisenberg1949, p.63]. Following this Heisenbergian interpretation, von Weizsäcker
[1985] and Mittelstaedt [1976] have construed C2 as a relation of mutual exclusiveness and joint completion between the abstract deterministic law of evolution of the
ψ-functions (i.e., the Schrödinger equation) and any measurement result observed
in space-time.
Because complementarity dismantles the articulation between the category of
causality and spatiotemporally-shaped sensory experience, and because this unsettles Kant’s system of a priori conditions of any possible knowledge, complementarity
appears to challenge not only the special architecture of the Critique of Pure Reason, but also, more generally, Kant’s global project of providing an account of how
the chaos of subjective impressions is progressively ordered into an objective pattern. As we saw in Section 2, the process of objectification involves, according to
Kant, two steps. In the first step, the impressions are embedded into a spatiotemporal structure, whereas in the second step, spatiotemporally located appearances are
connected with one another according to a law of succession. How can one objectify
the phenomena, if the former procedure is no longer available? Bohr’s answer is
daring, but in line with the spirit (if not the letter) of Kant’s epistemology. It consists in turning complementarity into a connecting device that supplants causality
(or rather includes causality within its own more general scheme of mutual exclusion and joint exhaustiveness). Quantum physics, Bohr says, ‘forces us to replace the
ideal of causality by a more general viewpoint’, namely ‘complementarity’ [Bohr1998,
p.84]. In so far as the standard (causal) connection of phenomena is prevented by the
16 [Kant1987,§59,
p.227],[Chevalley1995],[Pringe2007].
is for example a passage from the Como lecture [Bohr1934, p.54]: ‘. . . if in order to make observation
possible we permit certain interactions with suitable agencies of measurement, not belonging to the system, an
unambiguous definition of the state of the system is naturally no longer possible, and there can be no question
of causality in the ordinary sense of the word. The very nature of the quantum theory thus forces us to regard
the space-time co-ordination and the claim of causality, the union of which characterizes the classical theories, as
complementary but exclusive features of the description, symbolizing the idealization of observation and definition
respectively.’
17 Here
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
159
incompatibility of the various modes of experimental access, a new mode of (complementary) connection is offered as a substitute. ‘Complementarity is called for to
provide a frame wide enough to embrace the account of fundamental regularities
of nature which cannot be comprehended within a single picture.’ [Bohr1963, p.12]
As above, we can interpret Bohr’s move as aiming at extrapolating the mode of
existence of standard classic-like objects beyond the domain of ordinary experience.
The revolutionary mode of connection here compensates for a conservative ontology.
A diametrically opposed strategy has also been proposed, which consists in sticking
to the standard (causal) mode of connection at the cost of redefining the objects of
microphysics [Mittelstaedt1976].
5
Classical concepts: a remnant of Kant’s a priori?
Our last task is to explore the peculiar status of classical concepts within Bohr’s
philosophy, and to point out some functional analogies between Bohr’s classical
concepts and Kant’s a priori forms. The atom model [Bohr1913] provides a telling
example, from Bohr’s early career, of the role that classical concepts and representations played in his approach to theorizing. The atom model is a non-conventional
compromise between classical mechanics and electrodynamics on the one hand, and
Planck’s ‘quantum rules’ on the other. Because of its baroque combination of electronic stationary orbits ruled by classical mechanics, quantized transitions from one
orbit to another, and overt violation of certain theorems of classical electrodynamics,
the model has sometimes been regarded as incoherent. Yet, in so far as one follows
the prescriptions that restrict the use of the various pieces of this patchwork to
specific theoretical contexts, no inconsistency arises [Vickers2007]. In other terms,
the model lacks unity rather than logical and practical consistency. And unity, as a
regulative ideal of physics, was no sufficient motivation for Bohr to simply abandon
his clumsy model and endeavor toward a fully unified non-classical theory.
During the intermediate phase that separated the old quantum theory from
the advent of modern quantum mechanics, say from 1913 to 1925, Bohr was indeed looking for new ways of articulating classical physics with quantum postulates,
rather than trying to eliminate the classical features altogether. The ‘correspondence
principle’ between classical and quantum physics acted as a pivotal element of this
strategy. Not only was it taken as a prospective guide for the construction of new
theoretical structures (in a way that went far beyond the usual retrospective requirement that the old theory be a limiting case of the new one), but it was also used as
a sort of spare wheel for predicting the value of certain variables that did not appear in the hybrid model, such as the spectral line amplitudes. The correspondence
principle thus worked as a meta-theoretical structure that enabled one theoretical
structure (the classical one) to serve as analogic scaffolding for the development of
another theoretical structure (the quantum one) [Darrigol1992, p.81]:
. . . although the process of radiation cannot be described on the basis
of the ordinary theory of electrodynamics . . . there is found, nevertheless,
to exist a far-reaching correspondence between the various types of possible transitions between the stationary states on the one hand and the
various harmonic components of the motion on the other hand. . . . This
correspondence is of such a nature that the present theory of spectra is in
160
M. Bitbol and S. Osnaghi
Séminaire Poincaré
a certain sense to be regarded as a rational generalization of the ordinary
theory of radiation. [Bohr1922, pp.23-4]
Bohr’s effort to improve his original model involved a permanent negotiation
about which classical concepts should be retained and which should be excluded. The
boundary between the classical and the quantum components of the model kept moving accordingly. In an attempt to reconcile the discontinuity of the quantum jumps
with the continuous spatiotemporal orbits, in 1924 Bohr considered the possibility of
jettisoning the principle of causality and the principles of conservation of energy and
momentum for individual events [Bohr1924]. This proposal gained little support. In
a letter to Bohr of December 1924, Pauli suggested to reverse the approach, and to
dismiss the classical ‘picture’ of orbital trajectories in favor of a thoroughly ‘quantum’ account of the kinematic and dynamic aspects of the problem [Darrigol1992,
p.208]. A first sketch of such an account had just been provided by Born, who had
coined for it the name ‘Quantenmechanik (quantum mechanics)’ [Born1924]. The
next decisive accomplishment in this direction was, of course, Heisenberg’s [1925]
‘matrix mechanics’, which completely dispensed with the concept of a continuous
trajectory, and replaced it with a law-like structure that applied to the spectral observables. Although Heisenberg’s work implied a systematic procedure of symbolic
translation of classical laws into quantum laws, as well as the replacement of ordinary continuous variables by non-commuting matrices of measurable discontinuous
quantities, Bohr [1925] hailed it as a ‘. . . precise formulation of the tendencies embodied in the correspondence principle’. Heisenberg [1929] himself referred to his
theory as ‘a quantitative formulation of the correspondence principle’.18 The vestige
of classical physics was then far from having been eliminated.
The reasons for the stubborn presence of classical concepts within the new
theoretical structures might be sought in Bohr’s non-conventional approach to theorizing. This approach has cogently been compared to the methodology that Kant
discusses in his Critique of Judgment, which prescribes to bring seemingly heterogeneous theoretical structures into a unique system irrespective of their differences,
through a hypothetical use of reason [Pringe2009]. However, it would be misleading to ascribe Bohr’s motivation for retaining classical concepts to his methodology
of research alone. Neither had such a motivation anything to do with ontological
considerations. Rather, as Bohr himself explained in 1934, the crucial point was the
peculiar role that classical concepts play within language:
. . . it would be a misconception to believe that the difficulties of the
atomic theory may be evaded by eventually replacing the concepts of classical physics by new conceptual forms. . . . It continues to be the application
of these concepts alone that makes it possible to relate the symbolism of
the quantum theory to the data of experience. [Bohr1934, p.16]
Bohr’s emphasis on this argument came to occupy an increasingly important
place in his later writings. In 1949, he summarized it as follows:
However far the phenomena transcend the scope of classical physical explanation, the account of all evidence must be expressed in classical terms.
The argument is simply that by the word ‘experiment’ we refer to a situation where we can tell others what we have done and what we have
18 Quoted
in Darrigol [1992, p.276].
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
161
learned and that, therefore, the account of the experimental arrangement
and of the results of the observations must be expressed in unambiguous
language with suitable application of the terminology of classical physics.
[Bohr1958, p.39]
Bohr’s explicit reference to the conditions of possibility of communication contributes to further clarifying the relation between phenomenon and context. In Section 3, we saw that the transcendental status that Bohr attributed to classical concepts was linked to their role in implementing the separation between the measuring
instrument and the measured object, in a context in which such a separation appears to be precluded by the ‘essential wholeness’ resulting from the measurement
interaction [Bohr1958, p.72]. Since the object-instrument separation is, for Bohr,
demanded by the very concept of observation, classical concepts could be said to be
essential in order to make observation possible. This argument, however, is not devoid of ambiguities, since Bohr does not clearly (or at least, not always) distinguish
the semantic and the dynamical aspects involved in it.
In the preceding quotations, however, all the emphasis is placed on the conditions for communication. This should come as no surprise, given Bohr’s welldocumented interest in language. Bohr was always much concerned not only with
clarifying and circumscribing the rules for the valid use of particular concepts, as in
the atom model, but also with reaching a deeper understanding of the conditions for
objective description in general.19 His scattered remarks cannot of course be taken
as outlining, or presupposing, a definite theory of concepts. Yet, some general ideas
emerge if one considers his reflections as a whole (and in particular the post-EPR
papers). When adopting this standpoint, Bohr’s classical ‘conceptual framework’
may be thought as a system of ‘conceptual’ inferences which is presupposed by empirical inferences, in order for experience to have the constitutive features that we
ascribe to it. While the cut can accordingly be understood as a way to stress the
transcendental role of the classical conceptual framework, complementarity appears
as a (rather rudimentary) attempt to generalize such a framework, in a context in
which intrinsically probabilistic (i.e., non-causal) inferences are allowed.
Why, then, must we rely specifically on classical concepts in order to ‘tell others
what we have done and what we have learned’ ? Why, in other words, should ordinary language and classical physics (which we use to express experimental results
and to provide instructions on how to build and calibrate measuring instruments)
embody the most elementary conditions of possibility of intersubjective agreement in
general ? It might be thought that Bohr is here dogmatically assuming a positivistic
theory of meaning, according to which theoretical statements only have meaning in
so far as they can be translated into observational reports that refer ‘directly’ to the
objects of our immediate environment. Nothing, however, would be more foreign to
Bohr’s conception of language than the idea of privileging a class of ‘atomic’ propositions, based on the alleged immediacy of the link between those propositions and
their putative referents. What motivates Bohr’s prescription is, arguably, quite the
opposite view, namely that linguistic expressions acquire meaning only as part of a
web of inferential relations. The classical conceptual framework is in no way more
‘objective’ than other conceivable, and more general, systems of inferences. Yet, by
19 [Petersen1985]. See Murdoch [1987, Ch. 7] for a discussion of the pragmatist features in Bohr’s views on meaning,
particularly in so far as his partial endorsement of verificationist criteria is concerned.
162
M. Bitbol and S. Osnaghi
Séminaire Poincaré
instantiating the logical two-valued structure required for unambiguous communication, it makes objectivity possible in the first place.
6
Conclusion
In the Critique of Pure Reason, which addressed our interest in nature, Kant explained how, without truly disentangling ourselves from what there is, we can nevertheless elaborate a form of knowledge that works as if we were separated from
nature. Our role in this picture mimics that of an external spectator of nature, and
the knowledge thus acquired qualifies as objective. By contrast, in the Critique of
Practical Reason, which deals with the issues of freedom, action, and morals, any
such separation is precluded and we are ascribed the role of true actors of our own
deeds [Beck1963, p.31]. This dialectic of actor and spectator was later taken over
by Schopenhauer in The World as Will and Representation. According to Schopenhauer, the will is experienced from the point of view of a living actor, whereas
the representation of the world is obtained from a point of view that superficially
resembles that of a spectator. But the latent lesson of both Kantianism and postKantianism is that there is no true spectator’s standpoint (except in a minimalist
‘as if’ sense). And that in view of our insuperable entanglement with what there
is, the standpoint of a spectator of nature is extrapolated out of the only available
standpoint, which is that of the actor.
This is exactly what Bohr concluded from his reflection on quantum mechanics,
with some additional radicality though. He soon became aware that our apparent
disentanglement from nature in classical knowledge is only the limiting case of a
more general situation, in which it is not even possible to conceive of ourselves as
if we were spectators. This, however, does not entail that we can entirely dispense
with the role of the classical pseudo-spectator, if we are to make sense of experience
qua objective knowledge. ‘We are both onlookers and actors in the great drama of
existence’: this is ‘the old truth’ of which ‘the new situation in physics . . . has so
forcibly reminded us’ [Bohr1934, p.119].
References
[Allison2004] Allison, H.: Kant’s Transcendental Idealism. Revised and Enlarged
Edition. Yale University Press, 2004.
[Beck1963] Beck, L. W.: A Commentary on Kant’s Critique of Practical Reason.
The University of Chicago Press, 1963.
[Bell1990] Bell, J.: Against ‘measurement’. Physics World 8, 33–40 (1990).
[Bell2004] Bell, J.: Speakable and Unspeakable in Quantum Mechanics. Second Edition. Cambridge University Press, 2004.
[Bitbol1996a] Bitbol, M.: Schrödinger’s Philosophy of Quantum Mechanics. Kluwer,
1996.
[Bitbol1996b] Bitbol, M.: Mécanique quantique, une introduction philosophique.
Flammarion, 1996.
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
163
[Bitbol1998] Bitbol, M.: Some steps towards a transcendental deduction of quantum
mechanics. Philosophia naturalis 35, 253–280 (1998).
[Bitbol2010] Bitbol, M.: Reflective Metaphysics: Understanding Quantum Mechanics from a Kantian Standpoint. Philosophica 83, 53–83, (2010).
[Bohm1952] Bohm, D.: A Suggested Interpretation of the Quantum Theory in Terms
of ‘Hidden Variables’, I and II. Physical Review 85, 166–193 (1952).
[Bohm1993] Bohm, D., & Hiley, B.: The undivided universe. Routledge, 1993.
[Bohr1913] Bohr, N.: On the constitution of atoms and molecules. Philosophical
Magazine 26, 1–25 (1913).
[Bohr1922] Bohr, N.: The Theory of Spectra and Atomic Constitution. Cambridge
University Press, 1922.
[Bohr1924] Bohr, N., Kramers, H., & Slater, J.C.: The quantum theory of radiation.
Philosophical Magazine 47, 785–822 (1924).
[Bohr1925] Bohr, N. : Atomic theory and mechanics. Supplement to Nature 116,
845–852 (1925).
[Bohr1934] Bohr, N.: Atomic Theory and the Description of Nature. Cambridge
University Press, 1934.
[Bohr1958] Bohr, N.: Atomic Physics and Human Knowledge. John Wiley, 1958.
[Bohr1963] Bohr, N.: Essays 1958–1962 on Atomic Physics and Human Knowledge.
John Wiley, 1963.
[Bohr1998] Bohr, N.: Causality and Complementarity. The Philosophical Writings
of Niels Bohr, Volume 4. Ox Bow Press, 1998.
[Born1924] Born, M.: Über Quantenmechanik. Zeitschrift für Physik 26, 379–395
(1924).
[Brock2003] Brock, S. : Niels Bohr’s Philosophy of Quantum Physics. Logos Verlag,
2003.
[Bunge1955] Bunge, M.: Strife about Complementarity. British Journal for the Philosophy of Science 6(21), 1–12 (1955).
[Camilleri2009] Camilleri, K.: Constructing the Myth of the Copenhagen Interpretation. Perspectives on Science 17(1), 26–57 (2009).
[Cassirer1956] Cassirer, E.: Determinism and Indeterminism in Modern Physics.
Yale University Press, 1956.
[Cassirer1965] Cassirer, E.: Philosophy of Symbolic Forms. Yale University Press,
1965.
[Chevalley1991] Chevalley, C.: Le dessin et la couleur. In: N. Bohr. Physique atomique et connaissance humaine. Gallimard, 1991.
164
M. Bitbol and S. Osnaghi
Séminaire Poincaré
[Chevalley1995] Chevalley, C.: On objectivity as Intersubjective Agreement. In: L.
Krüger & B. Falkenburg (eds). Physik, Philosophie und die Einheit der Wissenschaften. Für Erhard Scheibe (pp. 332–346). Spektrum Akademischer Verlag,
1995.
[Cuffaro2010] Cuffaro, M.: The Kantian Framework of Complementarity. Studies in
History and Philosophy of Modern Physics 41, 309–317 (2010).
[Daneri1962] Daneri, A., Loinger, A., & Prosperi, G. M.: Quantum theory of
measurement and ergodicity conditions. Nuclear physics 33, 297–319 (1962).
Reprinted in: J. A. Wheeler & W. H. Zurek (eds.). Quantum theory and measurement (pp. 657–679). Princeton University Press, 1983.
[Darrigol1992] Darrigol, O.: From c-Numbers to q-Numbers. University of California
Press, 1992.
[Everett1957] Everett III, H.: ‘Relative State’ formulation of quantum mechanics.
Review of Modern Physics 29, 454–462 (1957). Reprinted in: J. A. Wheeler &
W. H. Zurek (eds.). Quantum theory and measurement (pp. 315–323). Princeton
University Press, 1983.
[Faye1991] Faye, J.: Niels Bohr, his Heritage and Legacy. Kluwer, 1991.
[Faye2010] Faye, J.: Niels Bohr and the Vienna circle. In: J. Manninen & F. Stadler
(eds.). The Vienna Circle in the Nordic Countries (pp. 33–45). Springer Verlag,
2010.
[Folse1978] Folse, H.: Kantian aspects of complementarity. Kant-Studien 69, 58–66
(1978).
[Folse1985] Folse, H.: The Philosophy of Niels Bohr: The Framework of Complementarity. North-Holland, 1985.
[Friedman1992] Friedman, M.: Kant and the Exact Sciences. Harvard University
Press, 1992.
[Fuchs2000] Fuchs, C., & Peres, A.: Quantum theory needs no ‘interpretation’.
Physics Today 53, 70–71 (2000).
[Jammer1974] Jammer, M.: The Philosophy of Quantum Mechanics. John Wiley,
1974.
[Joos1974] Joos, E., Zeh, H. D., Kiefer, C., Giulini, D., Kupsch, J., & Stamatescu, I.O.: Decoherence and the Appearance of a Classical World in Quantum Theory.
Springer, 2003.
[Kaiser1992] Kaiser, D. : More roots of complementarity: Kantian aspects and influences. Studies in History and Philosophy of Science 23, 213–239 (1992).
[Kant1987] Kant, I.: Critique of Judgment. Hackett, 1987.
[Kant1994] Kant, I.: Prolegomena to Any Future Metaphysics. Pearson, 1994.
[Kant1996] Kant, I.: Critique of Pure Reason. Hackett, 1996.
Bohr, 1913-2013, Vol. XVII, 2013
Bohr’s Complementarity and Kant’s Epistemology
165
[Kauark2012] Kauark-Leite, P.: Théorie quantique et philosophie transcendantale :
dialogues possibles. Hermann, 2012.
[Kuhn1962] Kuhn, T. K.: The structure of scientific revolutions. University of
Chicago Press, 1962.
[Heisenberg1925] Heisenberg, W.: Über die Quantentheoretische Umdeutung kinematischer und mechanischer Beziehunge. Zeitschrift für Physik 33, 879–893
(1925).
[Heisenberg1929] Heisenberg, W. : Die Entwicklung der Quantentheorie, 1918–1928.
Die Naturwissenschaften 17, 490–496 (1929).
[Heisenberg1949] Heisenberg, W.: The physical principles of the quantum theory.
Dover, 1949.
[Heisenberg1971] Heisenberg, W.: Physics and Beyond. Harper and Row, 1971.
[Heisenberg1990] Heisenberg, W.: Physics and philosophy. Penguin, 1990.
[Held1994] Held, C.: The meaning of complementarity. Studies in History and Philosophy of Science 25, 871–893 (1994).
[Held1995] Held, C.: Bohr and Kantian idealism. In: H. Robinson & G. Brittan
(eds.). Proceedings of the Eighth International Kant Congress (pp. 397–403).
Marquette University Press, 1995.
[Hermann1996] Hermann, G.: Les fondements philosophiques de la mécanique quantique. Vrin, 1996.
[Hintikka1991] Hintikka, J.: Dinge an sich revisited. In: J. Hintikka. Knowledge and
the Known (pp.197-211). Springer, 1991.
[Holton1978] Holton, G. J.: The Scientific Imagination: Case Studies. Cambridge
University Press, 1978.
[Honner1982] Honner, J.: The transcendental philosophy of Niels Bohr. Studies in
the History and Philosophy of Sciences 13, 1–30 (1982).
[Hooker1972] Hooker, C. A.: The nature of quantum mechanical reality. In: R. G.
Colodny (ed.). Paradigms and Paradoxes (pp. 135–172). University of Pittsburgh Press. 1972.
[Howard2004] Howard, D.: Who invented the ‘Copenhagen interpretation’ ? A study
in mythology. Philosophy of Science 71, 669–682 (2004).
[Mittelstaedt1976] Mittelstaedt, P.: Philosophical problems of modern physics. Reidel, 1976.
[Mittelstaedt1998] Mittelstaedt, P.: The Interpretation of Quantum Mechanics and
the Measurement Process. Cambridge University Press, 1998.
[Murdoch1987] Murdoch, D.: Niels Bohr’s Philosophy of Physics. Cambridge University Press, 1991.
166
M. Bitbol and S. Osnaghi
Séminaire Poincaré
[Omnès1992] Omnès, R.: Consistent interpretations of quantum mechanics. Reviews
of Modern Physics 64, 339–382 (1992).
[Osnaghi2009] Osnaghi, S., Freitas, F., & Freire, O.: The origin of the Everettian
heresy. Studies in History and Philosophy of Modern Physics. 17, 97–123 (2009).
[Pais1991] Pais, A.: Niels Bohr Time in Physics, Philosophy, and Polity. Oxford
University Press, 1991.
[Peres1982] Peres, A. & Zurek, W.: Is quantum theory universally valid? American
Journal of Physics 50, 807–810 (1982).
[Petersen1985] Petersen, A.: The Philosophy of Niels Bohr. In: A. P. French & P.
J. Kennedy (eds.). Niels Bohr. A centenary volume (pp. 299-310). Harvard
University Press, 1985.
[Pradelle2013] Pradelle, D.: Généalogie de la raison. P.U.F., 2013.
[Pringe2007] Pringe, H.: Critique of the Quantum Power of Judgment. A Transcendental Foundation of Quantum Objectivity. De Gruyter, 2007.
[Pringe2009] Pringe, H.: A transcendental account of correspondence and complementarity. In: M. Bitbol, J. Petitot & P. Kerszberg (eds.). Constituting Objectivity, Transcendental Perspectives in Modern Physics (pp. 317–327). Springer,
2009.
[Putnam1995] Putnam, H.: Pragmatism. Blackwell, 1995.
[Reichenbach1965] Reichenbach, H.: The Theory of Relativity and a priori Knowledge. University of California Press, 1965.
[Rosenfeld1965] Rosenfeld, L.: The measuring process in quantum mechanics. Supplement of the Progress of Theoretical Physics. 222–231 (1965).
[Rovelli1996] Rovelli, C.: Relational quantum mechanics. International Journal of
Theoretical Physics 35, 1637–1678 (1996).
[Schlosshauer2011] Schlosshauer, M., & Camilleri, K.: What classicality? Decoherence and Bohr’s classical concepts. Advances in Quantum Theory, AIP Conference Proceedings 1327, 26–35 (2011).
[Stapp1972] Stapp, H. P.: The Copenhagen Interpretation. American Journal of
Physics 40, 1098–1116 (1972).
[Vickers2007] Vickers, P.: Bohr’s Theory of the Atom: Content, Closure and Consistency. http://philsci-archive.pitt.edu/4005/ (2007).
[Weinberg2005] Weinberg, S.: Einstein’s mistakes. Physics Today 58 (11), 31–35
(2005).
[Weizsäcker1985] von Weizsäcker, C. F.: Aufbau der physik. Hanser, 1985.
[Zurek2003] Zurek, W. H.: Decoherence, einselection, and the quantum origins of
the classical. Reviews of Modern Physics 75, 715–775 (2003).